Integrating Probability- and Value Information in Judgment and Decision-Making under Risk : Cognitive Processes, Competence, and Performance

Sammanfattning: Many instances in human affairs involve considering the value of different outcomes and the probability (or risk) of these outcomes occuring (e.g., gambling, financial decision-making, medical decision-making, criminal behavior). The point of departure for present thesis is that descriptive theories of judgment- and decision making under risk have yet to fully utilize explanations grounded in accounts of how people integrate outcomes with their adherent probabilities. The most widely embraced accounts are positioned on opposite ends of a spectrum, holding either (i) that people consistently and effortlessly engage in the normative principle of multiplicatively integrating the value or utility of possible outcomes with their adherent probabilities (i.e., weighting), or (ii) that people only have the ability to engage in simple heuristics or context-dependent sampling strategies. The present thesis proposes that the field should consider positions between these extreme positions. To this end, three empirical studies were conducted in which people evaluated risky prospects in the form of numerically described monetary lotteries.The studies show that use of weighting was robust to increases of cognitive demands, as when (i) other evaluations are not available as reference points (Study I), (ii) outcomes and probabilities are presented sequentially before the evaluation (Study II), and (iii) the prospect structure involves two independent outcomes (Study III). The results suggest that - even if people can turn to heuristics when they are more efficient, for specific stages in the decision process, or for very complex problems - people indeed have both the inclination and ability to weight the outcomes by their probabilities in the evaluation of individual prospects, or for a subset of decision alternatives.In contrast to popular weighting models, however, the cognitive-modeling efforts throughout the studies speak against the notion that the weighting process can be assumed to be consistent and effortless. Instead, the cognitive process of weighting outcomes and probabilities is better characterized as an anchoring-and-adjustment strategy: people anchor on the value of the outcome and make linear adjustments downwards to account for probability. The studies show that these adjustments are often insufficient or noise-prone when the cognitive demands increase due to (i) properties of the task environment (Study I and Study II), or (ii) lack of domain-specific knowledge (i.e., numeracy and financial literacy, Study III). In conclusion, the thesis has highlighted the important, but previously neglected, nuances of human cognition in judgment and decision-making under risk - nuances found between previously conflicting standpoints. Future research exploring these nuances should make a necessary distinction between people’s underlying competence and the performance they exhibit at a given moment.